English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Sequence-to-sequence (seq2seq) learners are widely used, but we still haveonly limited knowledge about what inductive biases shape the way theygeneralize. We address that by investigating how popular seq2seq learnersgeneralize in tasks that have high ambiguity in the training data. We use SCANand three new tasks to study learners' preferences for memorization,arithmetic, hierarchical, and compositional reasoning. Further, we connect toSolomonoff's theory of induction and propose to use description length as aprincipled and sensitive measure of inductive biases.  In our experimental study, we find that LSTM-based learners can learn toperform counting, addition, and multiplication by a constant from a singletraining example. Furthermore, Transformer and LSTM-based learners show a biastoward the hierarchical induction over the linear one, while CNN-based learnersprefer the opposite. On the SCAN dataset, we find that CNN-based, and, to alesser degree, Transformer- and LSTM-based learners have a preference forcompositional generalization over memorization. Finally, across all ourexperiments, description length proved to be a sensitive measure of inductivebiases.

What they do when in doubt: a study of inductive biases in seq2seq learners